Existing-Asset AI Accelerant
Use proprietary data, products, and distribution to make AI more valuable.
- Difficulty
- Moderate
- Time to result
- ~months to results
- Steps
- 6
- Confidence
- 88%
The Existing-Asset AI Accelerant treats artificial intelligence as a multiplier rather than a standalone product. Start by identifying proprietary assets: unique data, established applications, trusted workflows, and distribution. Then apply AI where it makes those assets easier to search, combine, or act upon. Google illustrates the mechanism through products such as Gmail, Drive, Maps, Flights, Hotels, and YouTube: a conversational layer can coordinate information and actions that previously required several interfaces. The resulting advantage is cumulative. Proprietary data can improve relevance, integrated applications enable richer workflows, and an existing user base lowers adoption costs. The framework therefore directs strategy away from generic AI features and toward combinations that competitors cannot readily reproduce without the same underlying assets.
Origin
Extracted from Marketing Against The Grain during Kit Bodner and Kieran Flanagan's analysis of Google's advantages in the AI platform battle.
Core principles
- 01AI amplifies assets a company already owns.
- 02Proprietary data improves differentiation and defensibility.
- 03AI creates more value when embedded in familiar products.
- 04Existing distribution reduces the friction of adopting AI.
- 05Integrated workflows can outperform isolated AI tools.
How to run it
- 1
Inventory Existing Advantages
List the proprietary data, applications, workflows, customer relationships, and distribution channels the organization already controls. Distinguish genuinely scarce assets from information competitors can acquire easily.
Pro tip Prioritize assets that are frequently updated or embedded in recurring customer behavior.
Watch out Do not treat data as usable merely because the organization stores it; permissions and privacy constraints still apply.
- 2
Find High-Friction Workflows
Identify recurring tasks that force users to search manually, switch applications, reconcile information, or repeat instructions. Rank them by frequency, pain, and strategic importance.
Pro tip Look first for experiences where the existing search or navigation interface is notably weak.
Watch out Avoid adding AI to a workflow that is already faster and clearer without it.
- 3
Connect Data and Products
Give the AI controlled access to the relevant datasets and product capabilities so it can retrieve, combine, and transform information. Preserve source attribution and boundaries between services.
Pro tip Begin with read-only retrieval and synthesis before enabling consequential actions.
Watch out Broad access without careful authorization can expose private or irrelevant information.
- 4
Create a Unified Interface
Let users express the desired outcome in natural language instead of navigating each underlying application. The system should select and coordinate the required sources behind the scenes.
Pro tip Design prompts around complete user outcomes, such as planning a trip or finding every document about a subject.
Watch out A conversational interface should not hide uncertainty or prevent users from verifying results.
- 5
Use Distribution to Drive Adoption
Introduce the capability inside products customers already use rather than requiring them to adopt a separate AI destination. Demonstrate immediate improvements to familiar tasks.
Pro tip Lead with a noticeably better version of an existing behavior rather than abstract AI education.
Watch out Do not rely on distribution to compensate for an unreliable or unhelpful experience.
- 6
Compound the Advantage
Use observed workflows and feedback to improve retrieval, add integrations, and support progressively richer outcomes. Measure time saved, task completion, retention, and trust.
Pro tip Expand from one proven workflow into adjacent tasks that reuse the same assets.
Watch out Do not add integrations faster than the organization can govern and support them.
In the wild
Kieran describes asking Bard to search his Google Workspace information and construct a table containing payment dates, amounts, and contextual details for a company that pays him for advisory work. The AI turns scattered personal business records into a structured answer without requiring manual searches across documents and email.
→ An existing productivity suite becomes a faster financial-information retrieval and organization tool.
The hosts discuss an example in which Bard finds dates proposed by a friend, searches for flights on those dates, and builds an itinerary. The workflow combines personal information with travel products through one conversational request.
→ Several disconnected planning steps collapse into a coordinated outcome.
Kit proposes combining the show's top-performing episode data with calendar records of recording times. Bard could compare performance with production timing and reveal whether episodes recorded during a particular period tend to perform better.
→ Existing analytics and calendar data produce a testable operational insight.
Common mistakes
Building a Generic AI Wrapper
Adding the same model and features available to every competitor produces little defensibility. The advantage comes from combining AI with assets others cannot readily duplicate.
Ignoring Permissions
Ownership or access does not automatically justify every use of customer, community, or platform data. Unclear rights can turn a strategic asset into a legal and trust liability.
Forcing a Separate AI Habit
Requiring customers to leave familiar workflows increases adoption friction. Embedding AI into an existing product can make the transition feel like a natural improvement.
Is it for you?
Best for
It is best for organizations that already own valuable datasets, widely used products, or established customer relationships.
Not ideal for
It is not ideal for companies without differentiated assets or permission to use the data required by their AI system.
From the transcript
“we've talked before that AI is an accelerant for existing companies and those companies are going to benefit most”
“Google already has these incredible products and AI just makes them so much better”
“I think that is how people get onboarded to AI is it just makes the thing they're doing already much better”
From the episode
Google’s Secret Weapon To Destroy OpenAI For Good (#159)